For the past 12 years, the Protein Structure
Initiative (PSI) has been hard at work determining the structure of proteins in an effort to increase understanding of their
biological and biomedical functions. Almost 6000 structures have been solved to date and approximately one third of these
proteins do not have an assigned function.
Recently, researchers at MCSG, a PSI Center for
High-Throughput Structure Determination under the leadership of Andrzej
Joachimiak, described an experimental
method that can be used to determine functions of unknown proteins (Shumilin et al., 2012).
This high-throughput approach involves soaking
protein crystals with a metabolite cocktail followed by crystallographic screening and isothermal calorimetry. In this
particular study, crystals of the apo-form of three proteins belonging to two families of unknown function were exposed to 87
natural metabolites. Ligand binding produced electron density detectable by X-ray diffraction. Subsequent
exposure of the crystals to structurally related ligands served to optimize binding. The ligand-bound protein structures were
solved using molecular replacement (e.g., 3RPZ, pictured right) and isothermal calorimetry was used to characterize the high
affinity interactions. As a result of this study, the binding substrates of one family of proteins were identified, leading to
the determination of the biological function of an orphan enzyme. In addition, progress was made in determining the function of
the second family. A few of the many advantages of this method are that it is high-throughput, can utilize a
small metabolite screening pool, and relies on materials and methods that are readily available as a result of structural
This breakthrough approach on the experimental side of high-throughput functional proteomics complements the many computational methods that have been developed by many PSI Centers, including MCSG. MCSG has long been engaged in the challenge of classifying proteins of unknown function using computational approaches. Some of these computational tools and methods are Gene3D, ProFunc, and GeMMA. Gene3D is a comprehensive resource containing domain and functional annotations for over 10 million proteins, making it a valuable tool for investigation of known functions and interactions that can be used to expand explorations to new proteins (Lees J, et al., 2010). Automatic structure-based prediction of protein function can be achieved using the ProFunc server (Laskowski RA, et al., 2005). GeMMA is a tool that can be used to automatically classify protein families of sequence at all levels without initial sequence alignment (Lee DA, et al., 2010). These tools and many others can be accessed through MCSG's Model Validation & Fold Function Analysis page (from the main page, navigate to Technology and make your selection).
Other PSI Centers have established servers to assist researchers in assigning function to their proteins. JCSG (Ian Wilson, PI) administers TOPSAN, a wiki-based collaborative annotation web database designed to disseminate information about structures solved at PSI Centers (Ellrot K, et al., 2011). MarkUs is a structure-based function annotation server that was developed by NESG (Guy Montelione, PI) in the lab of Barry Honig (Petrey D, et al., 2009). During the early years of PSI, Integrated Center for Structure and Function Innovation (ISFI) built ProLinks for determining functional linkages (Bowers PM, et al., 2004). For more information about functional annotation resources, please visit the Protein Structures, Sequences, and Function Hub of the PSI SBKB.